217719 - automate the font type identification of 'Symbol' font (used in MathML)

Status: RESOLVED WONTFIX

(Core :: MathML, enhancement)

Description

[Gérard Milmeister]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686) Gecko/20030508 Galeon/1.3.7
Build Identifier: Mozilla/5.0 (X11; U; Linux i686) Gecko/20030508 Galeon/1.3.7

After mozilla 1.5a worked really well on MathML
(gtk2, Xft) build, I tried the new mozilla 1.5b.
The rendering is mostly correct, however some
characters are wrong, where they were right in mozilla 1.5a:

≠ displays as an up arrow
· displays as an angled left bracket

there are a few others too.

Also I noticed that the i -> inf under a lim in display math
is too low.

Reproducible: Always

Steps to Reproduce:

Comment 1

[Gérard Milmeister]

Ok, false alarm!

I forgot to set
encoding.symbol.ttf = Adobe-Symbol-Encoding
encoding.symbol.ftcmap = mac_roman
in
fontEncoding.properties

However this is a pitfall that should be corrected.
Is it possible to set entries in fontEncoding.properties
and mathfont.properties in user preferences?

Comment 2

[rbs]

No, it is not possible to override fontEncoding.properties and the operator
definitions (in mathfont.properties) using user preferences. You can only
set/override the font.* definitions in mathfont.properties via prefs, but not
the operator.* definitions.

The option is to edit the files, as you did.

As you have read in the comment of fontEncoding.properties, the reason why it is
so is because of the need to differentiate between the Type1 version and the
TrueType version. Last time I wondered on the matter, it wasn't possible to
automate that because the type of the font isn't passed onto Mozilla by the X
font server.

Comment 3

[Jungshik Shin]

There might be a way to identify the fonttype, but that would take some tweaking
of the overall logic of nsFontMetricsXft.cpp ( giving up lazy 'loading' of
actual fonts  because we have to rely on freetype API instead of fontconfig API)
 that I'd rather avoid.   I just filed a bug at fontconfig bugzilla
(http://fontconfig.org/cgi-bin/bugzilla/show_bug.cgi?id=109).
I'm assigning this to myself.

Comment 4

[Frédéric Wang (:fredw)]

Karl, is this bug still valid?

Comment 5

[Frédéric Wang (:fredw)]

And what about bug 121148 and bug 46622?

Comment 6

[Karl Tomlinson (:karlt)]

Now that we use Unicode fonts for general characters, differentiating between fonts of family name "Symbol" is only necessary for parts of stretchy characters.  We use Symbol only on WINNT, where the MonoType fonts should be present.  If it turns out that the Adobe Symbol font causes problems there, then perhaps we should remove that from the list of fonts used for stretchy characters, relying if necessary on the glyph-scaling code.